Do not be lazy.
You should always analyse your games first by yourself, then, if possible, together with another player of equal or (ideally) greater strength. This way you will hopefully not only come to understand the defects in your own thinking, but also the disparity between your appreciation of the game and that of another.
Then, and only then, should you subject your games to the analysis of Stockfish or another engine or engines. To quote GM Alexander Kalinin from his Chess Training for Candidate Masters:
"Many young chess players see the computer as the ultimate response to nearly everything. They think that computer analysis is the best and the fastest way to find the truth in any position on the board. As a result, many of those players have gradually stopped thinking and analysing for themselves."
In addition to this, there are many facets of human chess where a chess engine cannot help you. To pick just one, prophylaxis: the art of anticipating and frustrating your opponent's plans before they're allowed to materialise on the board. The chess engine, as a brute force calculating machine, has less need of such a concept than a human being. Humans must, to a certain extent, think in abstract terms, take necessary shortcuts, in order to compensate for the relative deficiency in our ability to calculate. Chess engines neither need nor are able (owing to a phenomenon called 'the horizon effect') to do this. Consequently, the engine will often ignore or devalue prophylactic ideas that a GM might applaud. Remember: the chess player aspires to be a better chess player, not a chess engine. One is possible, the other is not.
The absolute 'do not' method of analysing a chess game is to merely allow one of the popular chess-playing websites to do it for you. Whatever the drawbacks of going through the game with an engine alone, the additional analysis supplied by these websites, consisting mostly of '?!', '?' and other punctuation marks, is liable to deceive you as to all manner of truths regarding the practical quality of your play.
Don't take my word for this. (As Tarrasch said, the most important quality of a chess player is mistrust.) Test this assertion yourself by picking ten random Magnus Carlsen games and subjecting them to such analysis. You will discover that sometimes moves that the algorithm deems inaccurate are the beginnings of a decisive turn in his favour.
Analyse your games.
Write them up.
Seek to understand for yourself.
Then seek human assistance.
Then ask a computer.
Do not be lazy.